How to ignore Dropout and BatchNorm in test mode?


#1

In pytorch, I can use “mod.eval()” to set to the evaluation mode,Dropout and BatchNorm layers will be disabled then. How to do this in Mxnet, help please!


#2

See autograd help: http://mxnet.incubator.apache.org/api/python/autograd/autograd.html

you can use with mx.autograd.predict_mode() to control this